Abstract: A nonstationary Poisson model is proposed based on the characteristics of acceleration activity of earthquake occurrence in a seismic activity period. The seismic occurrence rate ν in the model increases exponentially with time, comparing with that the occurrencerate is a constant in stationary Poisson model. The log-likelihood method is used, and the parameters of the model can be obtained by maximizing its function. The authors discuss the difference between nonstationary Poisson model and stationary Poisson model by an assumed temporal sequence. The model is also used to the medium-and long term earthquake prediction along Fenwei seismic zone and North China plain seismic zone in North China.

Key words: Nonstationary Poisson model, Log-likelihood method, Medium-and long-term prediction, Historical earthquakes in North China